=Paper=
{{Paper
|id=Vol-1275/paper1
|storemode=property
|title=Okkam Synapsis: A Community-driven Hub for Sharing and Reusing Mappings Across Vocabularies
|pdfUrl=https://ceur-ws.org/Vol-1275/swcs2014_submission_1.pdf
|volume=Vol-1275
|dblpUrl=https://dblp.org/rec/conf/semweb/BortoliBB14
}}
==Okkam Synapsis: A Community-driven Hub for Sharing and Reusing Mappings Across Vocabularies==
https://ceur-ws.org/Vol-1275/swcs2014_submission_1.pdf
Okkam Synapsis: a community-driven hub for
sharing and reusing mappings across
vocabularies
Stefano Bortoli1 , Paolo Bouquet12 , and Barbara Bazzanella2
1
Okkam SRL
Via Segantini 23, I-38121 Trento, Italy
2
University of Trento - DISI
Via Sommarive, 14 I-38123 Povo di Trento, Italy
bortoli@okkam.it, bouquet@disi.unitn.it, barbara.bazzanella@unitn.it
Abstract. In the past 10-15 years, a large amount of resources have
been devoted to develop highly sophisticated and effective tools for au-
tomated and semi-automated schema-vocabulary-ontology matching and
alignment. However, very little effort has been made to consolidate the
outputs, in particular to share the resulting mappings with the commu-
nity of researchers and practitioners, support a community-driven re-
vision/evaluation of mappings and make them reusable. Yet, mappings
are an extremely valuable asset, as they provide an integration map for
the web of data and the “glue” for the Global Giant Graph envisaged
by Tim Berners-Lee. Aiming at kicking-off a positive endeavor, we have
developed Synapsis, a platform to support a community-driven lifecy-
cle of contextual mappings across ontologies, vocabularies and schemas.
Okkam Synapsis offers utilities to load, create, maintain, comment, sub-
scribe, and define levels of agreement over user defined contextual map-
pings available also through REST services.
Acknowledgement. This work is partially supported by TAG CLOUD
(Technologies lead to Adaptability and lifelong enGagement with culture
throughout the CLOUD) FP7 EU Funded project, Grant agreement nr:
600924.
1 Introduction and Motivation
In the promising vision of the Semantic Web proposed by Tim Berners-Lee [2],
the collaborative and distributed creation of semantically annotated documents
would enable software agents to perform time-consuming activities on behalf of
human users (see [1]). The community that gathered to corroborate and develop
this ambitious vision achieved many relevant results with the definition of impor-
tant standards such as OWL[19, 18, 14], RDF[15], and the important Linked Data
publication principles [3, 4]. The combination of these principles with the more
recent open data initiative across many countries is generating a considerable
amount of publicly available RDF and OWL data. In recent years, enterprises
are attracted by the promise of using such big and rich data to develop new prod-
ucts and services for their customers (e.g. [7]). However, exploiting and mining
data rises many challenges including the problems of entity matching, ontology
matching, and making the data accessible and usable by non-expert users. In the
past ten years many efforts were spent in the definition of sophisticated tools for
automated ontology matching. These often provided very effective solutions in
narrow domains, but a generic automatic reliable solution to the problem is still
an open research problem [20]. Furthermore, in [26] it was recently discussed how
often even experts have problems in finding agreement on defined ontology map-
pings. We argue that this is due to two main problems: 1) the intrinsic complexity
and heterogeneity of existing ontologies, and 2) the inconsistency and fuzziness
in usage of such ontologies due to contextually interpretable semantics. Namely,
concepts and relations expressed in natural language are interpreted outside the
original context of definition, and therefore prone to contextual interpretation.
In fact, besides the effort of researchers in formal ontology [13, 11, 10] the pro-
cess of ontology definition is driven by specific domain requirements and often
ontology engineering practices are neglected [16].
Under these premises, we decided to take one of the ten challenges of on-
tology matching described in [23] and confirmed in [20], and propose a novel
platform to support a collaborative ontology mappings definition and reuse [27].
The idea to take this challenge is rooted in the pragmatic need of resolving
the problem of semantic heterogeneity affecting a knowledge-based solution of
the entity matching in the context of the Semantic Web [5]. In particular, in
this work we argue that collecting and maintaining ontology mappings as con-
textual bridge rules [6] to harmonize the semantic of entities’ attributes can
provide great benefits by enabling the application of knowledge-based solution
to an entity matching problem [5]. Therefore, in our attempt to solve the entity
matching problem in the linked data, we produced several thousands of mappings
from existing ontologies, schemas and vocabularies towards a target ontology we
named Identification Ontology3 ([5] Chap. 5). Often these mappings were pro-
duced without considering the original, or intended, semantic of the properties,
but rather relying on its actual function looking directly into the data. This ap-
proach, besides being practical and concrete, interprets the ontology mappings
as contextual analogies as suggested in [21]. Namely, when producing mappings,
rather than considering the similarity among original intended functional pur-
pose of the properties (homology), we consider also its real function (analogy)
so that the mapping relation holds primarily on the instances level. On the one
hand, we are aware that this approach will create mappings that might not be
absolutely coherent and correct across several contexts, but as long as they serve
the purpose we can live with this limitation. On the other hand, we want to use
a first core set of mappings to kick-off a positive endeavor for the definition
of a platform to support a community-driven lifecycle of contextual mappings
3
http://models.okkam.org/identification_ontology.owl
between ontologies, vocabularies and schemas that could serve the definition of
new applications exploiting open linked data.
In this work we describe Okkam Synapsis, a web application conceived to
support the linked data community in creating, sharing and reusing contextual
ontology mappings to support the creation of novel services based on the linked
data consumption. Okkam Synapsis offers utilities to load, create, maintain,
comment, subscribe, and manage levels of agreement over user defined contex-
tual mappings. Most importantly, endorsing the recommendations described in
[26], we support different fine-grained typing models for the definition of the
mappings (e.g. OWL and SKOS) and compute level of agreement according to
different metrics to support filtering based on them. The mappings produced will
be available also through REST services, providing several levels of selection to
support diverse and unforeseen application scenarios. The purpose of the appli-
cation is to enable the users of Okkam Synapsis to collaborate in the definition
of mappings, commenting, rating, and subscribing them. Furthermore, we want
to allow users to explicitly define the context of use of the defined mappings, so
that others can take informed decision about reusing.
The underlying assumptions are:
– real linked data is in general too messy to rely on a unique set of mappings
in different contexts of use
– linked data may change in time, therefore contextual mappings must be
subject to specific lifecycle
– the number of existing vocabularies is growing, but reuse practices make the
manual mapping process feasible (see Linked Open Vocabulary4 )
– perfect agreement about defined mappings is unlikely to happen [26], better
let users to select what they need
The reminder of the paper is organized as follows: in Section 2 we overview
the related works dealing with crowd-sourcing of ontology mappings, and other
community-driven approaches; in Section 3 we describe in detail the platform,
discussing functions and services. In Section 4 an overview of the architecture of
the application is provided and finally in Section 7 we describe future work and
outline some concluding remarks.
2 Related Work
According to the most recent survey we are aware of [20], there are not many
tools supporting collaborative creation of ontology mappings. In [27] is described
a system for community-driven ontology matching, embedding provenance, fresh-
ness and other metadata suitable for the selection of the mappings. Besides the
a low-resolution screenshot presented in the paper, the system does not seem to
be available anymore. In [17] Noy et al. describe a system for the collection of
biomedial ontologies supporting the definition of mappings among them. Having
4
http://lov.okfn.org/dataset/lov/
collected more than 30.000 mappings, the authors propose a systems for filtering
and searching mappings. Furthermore, they argue about the concept of mappings
as bridges, and outline the need of specifying the type of relations (e.g. equiva-
lence). Currently the system is up and running, serving more than 370 biomedial
ontologies and several million of concepts. In [22] is described CrowdMap, a so-
lution for ontology matching based on crowd-sourcing. The ontology matching
task are decomposed in micro-tasks and submitted to workers of crowd-sourcing
platforms such as CrowdFlower and MTurk for manual evaluation. The results
obtained were compared with the one of automatic tools showing the feasibility
of the process. In [8] the authors discuss about the need of managing and reduc-
ing uncertainty related to crowdsourcing of ontology matching tasks, proposing
different ways to create micro-tasks suitable to increase possible agreements.
In [7] the authors describe Helix as a tool for creating ontology mapping as a
pay-as-you-go task while consuming linked data. In [12] the Correndo and Alani
describe OntoMediate, a project of the University of Southampton aiming at
supporting, among other functions, the creation and sharing of ontology map-
pings. Unfortunately, the project is over and to the best of our knowledge there
is no service available. Another trend in managing collective ontology matching
is through gamification. In [16] and [24] are described Guess What?! and Spot-
TheLink proposing the solution of ontology matching tasks in form of games to
give incentives and foster engagement to ease the cognitive effort of users and
stimulate the creation mappings and links in the linked data cloud. Noticeably,
to the best of our knowledge these systems are not currently available. In this
context we do not consider papers presenting automatic solutions to the ontology
matching for which we refer to the aforementioned survey [20].
In light of the analysis presented, to the best of our knowledge, the only sys-
tem available providing the services comparable with the one of Okkam Synapsis
is BioPortal [17]. However, given the vertical purpose of BioPortal and the lim-
ited collaborative features, we can safely affirm that there is room for a solution
such as the one proposed in this paper.
3 User Interface and Features
The current version of Synapsis distinguishes between two kinds of users: ad-
ministrators and end users. Administrators are users that have unrestricted ac-
cess to all the user-level functions, including uploading a source ontology, cre-
ating new mappings for concepts and properties, deleting existing mappings,
setting/changing the status of defined mappings, evaluating existing mappings
and reusing/exporting mappings. End users have only access to social functions
to express their level of agreement on previously created mappings and reusing
them. They can endorse and comment existing mappings, follow mappings they
are interested in, rate mappings and export mappings.
Figure 1 shows a snapshot of the User Interface of Okkam Synapsis which
presents three main areas: the (target) ontology on the left, the mappings in the
central part and the mapping filters on the right.
Fig. 1. Synapsis User Interface
After logging in, the user can select one of the ontologies/vocabularies cur-
rently present in the platform from the drop-down menu on the top-left corner
of the page or import a new ontology selecting the Import function from the
Function button. Following [17], we call the selected/uploaded ontology the Tar-
get Ontology5 , which is the ontology whose concepts/properties the user wants
to map to target concepts/properties. After having selected it, the target on-
tology is loaded, processed and represented as an indented tree on the left side
of the interface. The choice of using an indented tree is based on the study de-
scribed in [9], where users evaluated this representation model as easier to use
and more understandable than alternative models such as the graphs. With the
primary objective of enabling users in defining mappings, we decided to flatten
the ontology to a list of concepts and present the properties attached to them. In
the current version, we rely on a simple RDF processor implemented relying on
Apache Jena API6 . The selection of a node of the source ontology triggers the
loading of all the mappings defined for that concept or property in the central
part of the window. Each mapping is composed by the following attributes:
– Resource URI: the URI of the resource mapped towards the element of the
target ontology.
– Relation Type: the type of relation between the Resource URI and the Target
URI. The user can select among an enumeration of relations types including
5
According to this naming convention, a mapping can be seen as a relationship be-
tween two concepts/properties in different ontologies. Each mapping has a source
concept/property, a target concept/property, and a mapping relationship.
6
https://jena.apache.org/
OWL meta-relations such as owl:EquivalentProperty, owl:EquivalentClass,
owl:SubClass, owl:SubProperty and SKOS meta-relations skos:exact, skos:close,
skos:broader, skos:narrower. In this context, we neglect skos:related and
skos:unrelated because we believe these types of relations are not interesting
in our context. In order to help the user in choosing the right type of rela-
tion we refer to the guidelines proposed in [26] and still available at [25] as
appendix A2.
– Status: a label among Raw, Edited, Closed, Accepted, declaring the status of
a mapping. These labels are assigned by administrators of Okkam Synapsis
keeping into consideration time and opinions expressed by the members of
the community.
– Author: the author of the mapping.
– Description: A description of the resource mapped possibly coming from
official documentation.
– License: a statement declaring the licensing model under which the mapping
is made available to the community.
– Agreement Metrics: every user is enabled in stating whether she/he agrees
or not with the proposed mapping. The level of agreement may be estimated
using different metrics as suggested in [26].
– Number of Watchers: any mapping can be watched by a member of the
community. Watching a mapping allows users to be notified about activities
concerning the mapping.
– Number of Likes: any mapping can be liked by a member of the community.
A like essentially implies an agreement and a subscription to possible events
related to the mapping.
– Comments: members of the community are enabled in commenting and dis-
cussing about a mapping. We foresee cases where people may ask for clari-
fications and argue about the validity of the mapping.
– Contextual Tags: any mapping is annotated with a set of tags which identify
fuzzy contexts of application of the mappings. These tags can be used to
search and filter mappings.
In figure 1, one can see a graphical representation of all the mappings about
the Location concept of the Identification Ontology. The first two graphical
elements of the interface describe the relation and the status. The, after the
URI of the mapping, one can see the author of the mapping, the whether the
mapping was watched and by how many users. Finally, we show the number of
people care about that specific mapping. Then, on the right side of the interface,
a user can filter mappings according to these main dimensions, and typing on
the top input field, can filter mappings relying on the namespaces or the local
part of the mappings URIs.
Clicking on each mapping, the user can visualize all the details about the
mapping in a specific detail page (as shown in Figure 2). This page allows to add
comments, rate, subscribe and add possible contextual tags in a collaborative
manner. Ratings are made on a 6-item scale including the following options:
approved (i.e. the source and target concepts both mean the same thing), broader
(the target concept should be a broader term than the source concept), narrower
(i.e. the target concept should be a more specific term than the source concept),
related (i.e. the two concepts are not an exact match but they are closely related),
not sure (i.e. there is a relationship between the two concepts but none of the
above relations are appropriate or the term is used in a confusing or contradictory
fashion), rejected (i.e. the two concepts are definitely not the same, nor do they
have any other direct relationship with each other as listed above). The mapping
detail page essentially aims to provide tools for the collaborative interaction for
each single defined mapping. If a user subscribes a mapping, any notification
will include a link to the specific mapping detail page.
Fig. 2. Mapping page
Once selected the target ontology, the user is enabled in filtering mappings
according to different features. On the right part of the page, the filter features
are displayed, and the user is enabled in selecting them. Each selection triggers
an action on the list of mappings, removing the filtered ones. Mappings can be
filtered by author name, status, relation type, rating and creation date. It is also
possible to select all the mappings that have comments.
4 Architecture and Data Model
The application is designed according to the traditional MVC design pattern,
relying on J2EE JSF framework7 for the Web interaction part. The mappings
7
http://docs.oracle.com/javaee/5/tutorial/doc/bnaph.html
are also available through rest service, which are implemented relying on Jersey
framework8 . The AJAX based user interface interaction grants quick response
and easy interaction for the user. A view of the architecture of the application
is presented in figure 3.
Fig. 3. A graphical view of the Synapsis architecture
Both the J2EE Backing Beans and the REST services access the mappings
through standard Data Access Objects (DAO), which rely on the Hibernate
ORM JPA provider9 to interact with a relational database containing all data
about mappings, users and supported models. A detailed view of the data model
underlying the database is presented in figure 4.
5 Licensing Model
Any mapping created and shared through Okkam Synapsis is realeased under
the very popular Common Creative Attribution Share-alike 4.0 International li-
cense10 (CC BY-SA 4.0). The adoption of this copy-left license is to guarantee
correct attribution and sharing without affecting the re-usability of the map-
pings (including commercial purposes). Therefore, contributors and consumers
of mappings are granted what we believe is the ideal level of flexibility to ac-
commodate requirements both of researchers and companies. Notice that all the
mappings created through the application as subject to this license. Future evo-
lution of the application may allow the selection of other licensing models to
be compliant with loading of batches of mappings created else-where and under
different licensing model including the share-alike.
8
https://jersey.java.net/
9
http://hibernate.org/orm/
10
http://creativecommons.org/licenses/by-sa/4.0/
Fig. 4. Synapsis Data Model
6 Kick-off Mappings Dataset
Currently, Synapsis stores 22 mapping for equivalent class, and 205 mapping
for subclasses of the entity type Person; 22 mappings for equivalent classes, and
2322 mappings for sub classes of the type Location; and finally we defined 20
mappings for equivalent classes and 2468 mappings for subclasses of the type
Organization. These mappings were generated as contextual bridge rules to sup-
port semantic harmonization tasks in the knowledge-based solution described in
[5]. In particular, the reader can find details about the process leading to the
creation of such mappings from existing vocabularies towards the Identification
Ontology11 in Chapter 7 of [5]. We believe that this first core set of mappings
can help to kick of a positive endeavor in the adoption of the Okkam Synapsis
as a platform to create, share and manage mappings among vocabularies.
7 Conclusion and Future Work
In this paper we have presented a platform called Synapsis which provides a
gateway to collaboratively-defined ontology mappings. Looking for a pragmatic
11
http://models.okkam.org/identification_ontology.owl ([5] Chap. 5)
solution to the real world heterogeneity, complexity and inconsistencies, we de-
cided enable users to define mappings as contextual bridge rules, and enable
peers to comment and discuss about them. We believe that rating and estima-
tion of level of agreement about mappings would allow to filter commonly shared
mappings, and at the same marginalize odd ones. A beta version of the applica-
tion is available at http://api.okkam.org/synapsis, and can be preliminarily
tested and evaluated. In the next future, we plan to extend the support for the
definition of mappings around applications, to support Linked Data application
developer to select the set of mappings of interest and have them available for
the application through the defined rest services.
References
1. G Antoniou and F. van Harmelen. A Semantic Web Primer. MIT Press, 2004.
2. T. Berners-Lee, J. A. Hendler, and O. Lassila. The Semantic Web. Scientific Amer-
ican, May, 2001. http://www.sciam.com/2001/0501issue/0501berners-lee.html.
3. Tim Berners-Lee. Design Issues – Linked Data. Published online, May 2007.
http://www.w3.org/DesignIssues/LinkedData.html.
4. C. Bizer, R. Cyganiak, and T. Heath. How to publish linked data on the web.
online tutorial, July 2007.
5. Stefano Bortoli. Knowledge Based Open Entity Matching. PhD thesis, International
Doctoral School in ICT of the University of Trento (Italy), 2013.
6. Paolo Bouquet, Fausto Giunchiglia, Frank Harmelen, Luciano Serafini, and Heiner
Stuckenschmidt. C-owl: Contextualizing ontologies. In Dieter Fensel, Katia Sycara,
and John Mylopoulos, editors, The Semantic Web - ISWC 2003, volume 2870 of
Lecture Notes in Computer Science, pages 164–179. Springer Berlin Heidelberg,
2003.
7. Jason B. Ellis, Oktie Hassanzadeh, Kavitha Srinivas, and Michael J. Ward. Col-
lective ontology alignment. In OM, pages 219–220, 2013.
8. Jrme Euzenat. Uncertainty in crowdsourcing ontology matching. In OM, pages
221–222, 2013.
9. Bo Fu, Natalya F. Noy, and Margaret-Anne Storey. Indented tree or graph? a
usability study of ontology visualization techniques in the context of class mapping
evaluation. In The Semantic Web ISWC 2013, pages 117–134. Springer Berlin
Heidelberg, 2013.
10. A. Gangemi and V. Presutti. Ontology Design Patterns, pages 221–243. Springer
Berlin Heidelberg, 2009.
11. Aldo Gangemi, Nicola Guarino, Claudio Masolo, Alessandro Oltramari, and Luc
Schneider. Sweetening ontologies with dolce. In Proceedings of the 13th Inter-
national Conference on Knowledge Engineering and Knowledge Management. On-
tologies and the Semantic Web, EKAW ’02, pages 166–181, London, UK, UK, 2002.
Springer-Verlag.
12. Harith Alani Gianluca Correndo. Collaborative support for community data shar-
ing. In Proceedings of The 2nd Workshop on Collective Intelligence in Semantic
Web and Social Networks, 2008.
13. Nicola Guarino and Chris Welty. An overview of ontoclean. In Steffen Staab and
Rudi Studer, editors, The Handbook on Ontologies, pages 151–172. Springer-Verlag,
2004.
14. Pascal Hitzler, Markus Kroetzsch, Bijan Parsia, Peter F. Patel-Schneider, and
Sebastian Rudolph. OWL 2 Web Ontology Language Primer (Second Edition).
W3C, December 2012.
15. Frank Manola, Eric Miller, and Brian McBride. RDF 1.1 Primer. W3C, w3c
working group note edition, June 2014.
16. Thomas Markotschi and Johanna Völker. Guess What?! human intelligence for
mining linked data. In Proceedings of the Workshop on Knowledge Injection into
and Extraction from Linked Data (KIELD) at the International Conference on
Knowledge Engineering and Knowledge Management (EKAW), 2010.
17. NatalyaF. Noy, Nicholas Griffith, and MarkA. Musen. Collecting community-based
mappings in an ontology repository. In Amit Sheth, Steffen Staab, Mike Dean, Mas-
simo Paolucci, Diana Maynard, Timothy Finin, and Krishnaprasad Thirunarayan,
editors, The Semantic Web - ISWC 2008, volume 5318 of Lecture Notes in Com-
puter Science, pages 371–386. Springer Berlin Heidelberg, 2008.
18. W3C OWL Working Group. OWL 2 Web Ontology Language: Document
Overview, 27 October 2009. Available at http://www.w3.org/TR/owl2-overview/.
19. P.F. Patel-Schneider, P. Hayes, and I. Horrocks. Web Ontology Language (OWL)
Abstract Syntax and Semantics. Technical report, W3C, February 2003. http:
//www.w3.org/TR/owl-semantics/.
20. Shvaiko Pavel and Jerome Euzenat. Ontology matching: State of the art and future
challenges. IEEE Trans. on Knowl. and Data Eng., 25(1):158–176, January 2013.
21. Elie Raad and Joerg Evermann. Is ontology alignment like analogy? – knowledge
integration with lisa. In Proceedings of Symposium On Applied Computing (SAC),
Korea, Republic Of (2014), 2014.
22. Cristina Sarasua, Elena Simperl, and Natalya F. Noy. Crowdmap: Crowdsourcing
ontology alignment with microtasks. In Proceedings of the 11th International Con-
ference on The Semantic Web - Volume Part I, ISWC’12, pages 525–541, Berlin,
Heidelberg, 2012. Springer-Verlag.
23. Pavel Shvaiko and Jérôme Euzenat. Ten challenges for ontology matching. In
Proceedings of the OTM 2008 Confederated International Conferences, CoopIS,
DOA, GADA, IS, and ODBASE 2008. Part II on On the Move to Meaningful
Internet Systems, OTM ’08, pages 1164–1182, Berlin, Heidelberg, 2008. Springer-
Verlag.
24. Stefan Thaler, Elena Simperl, and Katharina Siorpaes. Spotthelink: Playful align-
ment of ontologies. In Proceedings of the 2011 ACM Symposium on Applied Com-
puting, SAC ’11, pages 1711–1712, New York, NY, USA, 2011. ACM.
25. Anna Tordai. On Combining Alignment Techniques. PhD thesis, Vreije Universiteit
Amsterdam, 2012-12-03.
26. Anna Tordai, Jacco van Ossenbruggen, Guus Schreiber, and Bob Wielinga. Let’s
agree to disagree: On the evaluation of vocabulary alignment. In Proceedings of the
Sixth International Conference on Knowledge Capture, K-CAP ’11, pages 65–72,
New York, NY, USA, 2011. ACM.
27. Anna V. Zhdanova and Pavel Shvaiko. Community-driven ontology matching. In
Proceedings of the 3rd European Conference on The Semantic Web: Research and
Applications, ESWC’06, pages 34–49, Berlin, Heidelberg, 2006. Springer-Verlag.